Command line tool to validate and normalize UTF-8 data
Go to file
Leonora Tindall d16a929926 Update dependencies. This is 1.1.1! 2023-04-02 22:03:04 -05:00
src Add --buffered and same stream check 2019-10-21 08:26:35 -05:00
.gitignore First commit - working version from RBR 2019-10-19 16:05:52 -04:00
Cargo.lock Update dependencies. This is 1.1.1! 2023-04-02 22:03:04 -05:00
Cargo.toml Update dependencies. This is 1.1.1! 2023-04-02 22:03:04 -05:00
README Add examples and clarify documentation 2023-04-02 22:02:15 -05:00


utf8-norm, validate and normalize UTF-8 Unicode data


Version 1.1.0 licensed GPLv3. (C) 2019 Leonora Tindall <>
Fast command line Unicode normalization, supporting stream safety transformations as well
as NFC, NFD, NFKD, and NFKC. Exits with failure if the incoming stream is not valid UTF-8.


Usage: utf8-norm [--nfc | --nfd | --nfkc | --nfkd] [--stream-safe] [--crlf] <infile> <outfile>

    <infile> (default stdin) - file from which to read bytes.
    <outfile> (default stdout) - file to which to write normalized Unicode.
    -w, --crlf  - write CRLF (Windows) instead of LF only (Unix) at the end of lines.
    -d, --nfd   - write NFD (canonical decomposition).
    -D, --nfkd  - write NFKD (compatibility decomposition).
    -c, --nfc   - write NFC (canonical composition computed from NFD). This is the default.
    -C, --nfkc  - write NFKC (canonical composition computed from NFC).
    -s, --stream-safe   - write stream-safe bytes (Conjoining Grapheme Joiners, UAX15-D4).
    -b, --buffered  - read the entire input file into memory before operating on it.
    -V, --version - output version information and exit.

utf8-norm operates linewise on the input unless --buffered is specified.

The --buffered option is primarily useful for reading and writing to the same file. It will
read bytes from the input until end of file and only then begin processing lines of the


Write the contents of input.txt, compatibly decomposed, with CRLF line endings,
to output.txt:

utf8-norm --nfkd --crlf input.txt output.txt

Normalize, in the canonical composition, buffering the file in memory to
avoid overwriting it with zeros:

utf8-norm --buffered

Emit the output of my_program to stdout, in the canonical composition, linewise.

my_program | utf8-norm

Buffer the entire output of my_program in memory, and emit it to
my_program.output in the canonical composition after receiving end-of-file.

my_program | utf8-norm --buffered - my_program.out


utf8-norm was created at Rust Belt Rust 2019 in Dayton, OH. Thanks to @j41manning for her
excellent talk regarding Unicode handling in Rust.

Natively install as `cargo install utf8-norm` or from your distribution's package manager.